Excelero says that its NVMesh® server SAN now supports persistent container storage for hyperscale architectures utilising Kubernetes. This represents the industry’s first offering making use of pooled, redundant NVMe storage in container applications requiring persistent volumes, so enterprises can obtain both local flash performance and container mobility at data centre scale. Excelero presented the new capability at the Technology Live! event in London, a blogger and analyst-focused drill-down into key storage and virtualization technologies.
Click to Tweet: @ExceleroStorage adds persistent #container storage for #Kubernetes pods #K8s http://bit.ly/2s32pg4
Scalable and persistent container storage is a top pain point for many infrastructure architects and developers who want to embrace microservices style deployments but desire more performant approaches than NFS or persistent volumes on traditional all-flash-arrays (AFAs). They want the local performance of flash but the flexibility and data protection of centralised storage. The new NVMesh capability uses Kubernetes’ advanced orchestration layer to deliver pooled NVMe with local latency and performance. In this approach, containers in a pod can access persistent storage presented to that pod, but with the freedom to restart the pod on an alternate physical node.
"Container users need persistent storage that is scalable enough for stateful applications yet also offers mobility to help protect against drive or host failure,” said Mark Peters, practice director & senior analyst at ESG. “By leveraging Kubernetes with its server SAN platform, Excelero is enabling IT teams to have exactly that: containers that can have high performance storage with both persistence and mobility.”
Excelero’s NVMesh 100% Server SAN platform further benefits container deployments with its approach by shifting data services from centralised CPU to complete client-side distribution. It virtualises the NVMe devices and unifies the capacity into a single pool of high-performance storage in an approach that makes data locality irrelevant, a breakthrough in enabling local latency and speeds on the network, using standard hardware. Because NVMesh does not impose a “CPU tax” on targets sharing NVMe drives, it allows for complete converged deployments without the normal SDS penalty. This allows NVMesh to scale performance linearly at near 100% with a virtual, distributed non-volatile array without requiring additional dedicated storage servers or appliances.
“Robust, low-latency container storage is long overdue as more IT teams look to microservices requiring persistence to improve the speed and reliability of cloud environments,” said Josh Goldenhar, vice president of products at Excelero. “We’re committed to embracing leading-edge innovations including Kubernetes pods so that our customers can build IT architectures at data centre scale backed by more efficient Flash storage.”
